By Noah Hendelman
The 2024 election ushered a new cleavage to the forefront of American politics: gender. Reporting prior to the 2024 election forewarned of a gender gap unparalleled in modern politics, large enough to impact the election. Yet, exit polls indicate the 2024 election saw a 10 point gap between men and women who voted for President Trump — well in-line with the 7 to 12 point gender gap of the past seven Presidential elections.
While a lack of female support for Harris helps explain 2024’s precedented gender divide – women voting for the Democratic party declined by 4 percentage points between 2020 and 2024 – a staggering 55% of men voted for Trump, the highest share since 2004. Given that evidence of a male shift toward the Republican Party persists, hypotheses have been set forth to explain this shift in the male demography: economic malaise, cultural sidelining, podcasts. We look at a potential explanation, “the China Shock,” to better understand the accuracy of economic explanations behind shifting male political preferences.
The China trade shock describes the economic disruption caused by a surge in low-cost manufacturing imports from China since 2001. This disruption has disproportionately impacted male-dominated industries such as automotive, steel, and manufacturing. Beyond its economic effects, the shock carries a political dimension, as growing import competition from China has contributed to rightward shift in media consumption and gains in the number of conservative representatives at the cost of moderate democrats. Given that male employment is most affected by the shock, and the shock has advanced right wing politics, we ask whether there is a direct connection at the individual level between exposure to the China Shock and political affiliation.
To do so, we use 2022 data from the Harvard Cooperative Election Survey (CES) containing voter identification and party affiliation information. We use a random forest method to estimate probabilities for party identification (Democrat, Republican, and Independent). Random forests are ensembles of regression trees, which partition the data into subsets by sequentially splitting variables such as age, race, gender, and socioeconomic status based on their association with individual party identification. We then merge CES data containing party identification estimates with data from Autor, Dorn, and Hanson, quantifying economic exposure to imports from China during the period from 2000-2007, by commuting zone. Thus our dataset includes around 50,000 individuals, with measures of their ideology and exposure to the China shock. We regress predicted party identification on the China shock and an interaction between the shock and gender to understand whether the effect of the China shock on party identification differed by gender.
The shock estimates are calculated by dividing the change in Chinese imports for an industry by that industry’s initial total employment, generating per-worker import exposure measures in dollars. While women are more likely than men to identify as Democrats across all levels of the shock, the shock’s impact on their likelihood to identify as a Democrat is not statistically different compared to men. A $1000 increase in shock is associated with a 0.65 percentage point decrease Democratic Party identification among both genders, meaning the shock pushes men and women away from the Democratic party to a similar extent.
Many of those who leave the Democratic party identify as Independents. A $1,000 increase in China shock is associated with a 0.35 percentage point increase in the chance of identifying as an independent. The shock increases the likelihood of identifying as an independent to a similar degree regardless of gender.
Gender dynamics emerge when analyzing the shock’s effect on Republican ideology. Higher shock levels not only pull voters into the Republican party, but in opposition to our original hypothesis, the effect is larger for women. A $1000 increase in shock is associated with a 0.16 percentage point increase in the chance a male identifies with the Republican Party but a 0.43 point increase among women, a statistically significant difference. Living in an area that has experienced a large China shock shifts women rightward more than men.
This result is puzzling, given that significant portions of shock-induced job loss are centered around industries dominated by men. The results reaffirm how politics is informed by more than personal
experience in the labor market. The effects of economic insecurity on family or community, or subjective perceptions of the broader economy, can shape political ideology more than direct occupational experience. Our results indicate voters in trade-competing areas exhibit some political malleability, leaving the left at faster rates than they are joining the right or becoming independents. This dynamic creates an opportunity for coalition-building among parties willing to engage with those who have been left behind by globalization.
![]() |
Article by Noah Hendelman ’25 Senior Data Journalist Manager |
![]() |